Non-covalent molecular complexes play a crucial role in regulatory biological processes, such as, but not limited to gene expression, DNA replication, signal transduction, cell-to-cell interaction, and the immune response. The molecular mechanisms of action of many drugs are based on drugs forming non-covalent molecular complexes with therapeutic targets. In addition, the formation of non-covalent molecular complexes is pivotal to many analytical techniques and devices used in research and disease diagnostics, such as, but not necessarily limited to, immunoassays, biosensors, and DNA hybridization analyses (Cepek and Brenner, Nature 1994, 372, 190; Sparks et al. Med. Chem. 1993, 36; Cohen and Williams, Microbiol. Sci. 1988, 5, 265; Pantoliano and Horlick, Biochemistry 1994, 33, 10229; Karlsson, Trends Pharm. Sci. 1991, 12, 265; Dalgleish and Kennedy, Vaccine 1988, 6, 215; Christian et al. Biochem. J. 1994, 300, 165).
The formation and decay of a non-covalent complex, L·T, between molecules L (ligand) and T (target), are characterized by a bimolecular rate constant kon, and a monomolecular rate constant, koff, respectively:
where kon is the rate constant of the forward reaction forming L·T and koff is the rate constant of the reverse reaction. The stability of the complex is often described in terms of the equilibrium dissociation constant:Kd=koff/kon  (2)The three constants, kon, koff, and Kd, are interconnected through equation 2, therefore defining any pair of constants will define the third. The constants are also dependent on a number of parameters, such as but not limited to buffer composition, buffer pH, buffer ionic strength, and temperature.
Determination of kon, koff and Kd. Knowledge of kon, koff, Kd, and their dependence on certain factors such as buffer composition, buffer pH, buffer ionic strength, and temperature can assist in: (i) understanding the dynamics of biological processes, (ii) determining the pharmacokinetics of target-binding drugs, and (iii) the designing of quantitative affinity analyses. In practical terms the determination of kon, koff, Kd can assist in developing and/or selecting drugs with desired kinetic parameters. It may also help in developing suitable dosage regimes. In another aspect it can help in the development of screening and/or diagnostic assays.
Prior art methods for measuring kon, koff and Kd of a molecular interaction have significant limitations. The methods that are used for finding kon and koff can be divided into two broad categories: heterogeneous and homogeneous binding assays. In heterogeneous assays, T is affixed to a solid substrate, while L is dissolved in a solution and can bind T affixed to the surface. In advanced heterogeneous binding assays such as surface plasmon resonance (SPR), T is affixed to a sensor that can change its optical or electrical signal upon L binding to T (Imanishi and Sugiura, Biochemistry, 2002, 41, 1328; Cheskis and Freedman, Biochemistry, 1996, 35, 3309). In such methods, Kd can be found by performing a series of equilibrium experiments. The concentration of L in the solution is varied and the interaction between L and T is allowed to reach equilibrium. The signal from the sensor versus the concentration of L has a characteristic sigmoidal shape and Kd can be found from the curve by identifying the concentration of L at which the signal is equal to half of its maximum amplitude. The koff value can be determined by SPR in a single non-equilibrium experiment in which the equilibrium is disturbed by fast replacing the solution of L with a buffer devoid of L. The complex on the surface decays in the absence of L in the solution, and the complex decay generates an exponential signal on the sensor.
Heterogeneous binding assays have certain advantages and drawbacks. The most serious drawback is that affixing T to the surface changes the structure of T. The extent of such change will depend on the method of immobilization. The change in the structure can potentially affect the binding parameters of L to T. This problem is especially severe for L that binds to T through interaction with a large part of T. In addition, the immobilization of T on the surface may be time-consuming and expensive. Moreover, non-specific interactions with the surface are always a concern.
In homogeneous binding assays T and L are mixed and allowed to form a complex in solution; neither of the molecules are affixed to the surface. Complex formation is followed by either monitoring the changing physical-chemical properties of L or T upon binding. Such properties can be optical (absorption, fluorescence, polarization) or separation-related (chromatographic or electrophoretic mobility). Equilibrium experiments with varying concentrations of L can be used similarly to heterogeneous analyses to find Kd. Non-equilibrium stopped flow-experiments, in which L and T are mixed in a fast fashion and the change in spectral properties is monitored, can be used to find kon. Non-equilibrium chromatographic experiments, in which a competitive ligand is added to the chromatographic buffer and allowed to interact with T was demonstrated to be useful in finding Kd and koff, although the method involved “non-transparent” numerical analysis of chromatographic peaks and required an additional reactant, the competitive ligand.
When the quantity of available T or L is a limiting factor, capillary electrophoresis (CE) is the method of choice. It requires only nanolitre (nL) volumes of a sample and can detect fewer than 1000 molecules (Wu and Dovichi J. Chromatogr. 1989, 480, 141). Affinity capillary electrophoresis (ACE), in which L is added to the run buffer at different concentrations and the change of the mobility of T is monitored, can be used to determine Kd by conducting a series of equilibrium experiments (Wan and Le, Anal. Chem., 2000, 72, 5583; Le et al., Electrophoresis, 2002, 23, 903; Chu et al. J. Med. Chem., 1992, 35, 2915; Chu and Whitesides, J. Org. Chem. 1992, 57, 3524; Carpenter et al. J. Chem. Soc., Chem. Commun. 1992. 804; Chu et al. Cell. and Mol. Life Sci., 1998, 54, 663). However, ACE is an equilibrium approach that cannot be used for finding koff.
Quantitative affinity analyses. Equilibrium binding analyses described in the previous section can be converted into methods for the quantitative analysis of T using the affinity probe L. Three major categories of affinity probes include antibodies (used in immunoassays), DNA hybridization probes (used in analyses of DNA and RNA) (Pease et al. P. Natl. Acad. Sci. USA 1994, 91, 5022; Mullaart et al. Nature, 1993, 365, 469; Higuchi et al. Nature, 1988, 332, 543), and aptamers (synthetic affinity probes, e.g. oligonucleotides or oligopeptides) (Clark and Remcho, Electrophoresis 2002, 23, 1335; Li et al. Biochem. Biophys. Res. Commun., 2002, 292, 31; Fredriksson et al. Nat. Biotechnol., 2002, 20, 473). When a target is available in very low amounts and cannot be amplified (i.e. not subject to PCR), CE can be the method of choice for developing a quantitative affinity analysis (Colton et al. Electrophoresis, 1998, 19, 367; Anderson et al. Anal. Chem., 2002, 74, 1870; Tim et al., Electrophoresis, 2000, 21, 220; Heegaard et al., Electrophoresis, 1999, 20, 3122; Busch et al., J. Chromatography A, 1997, 777, 311). In CE-based quantitative analyses the mobility shift of the affinity probe, L, is measured upon binding to the target, T. The shift is a function of the concentration of T. One of the major limitations of CE is its poor performance for antibodies as affinity probes. High molecular weight of antibodies significantly limits the mobility shift upon binding to a usually smaller target molecule. Recent advances in developing oligonucleotide aptamers open the possibility of their use as affinity probes in CE-based analyses (German et al. Anal. Chem., 1998, 70, 4540). Such analyses have been demonstrated in ACE mode, where the target is added to the buffer. However, ACE has two drawbacks for quantitative affinity analyses. If the target is a protein, its addition to the separation buffer is typically associated with protein adsorption to capillary walls, which can severely affect the quality of analysis (Gomez et al., Anal. Chemistry, 1994, 66, 1785). Second, adding the target to the running buffer can be unacceptable if the amount of target available is very small. An alternative to the ACE analysis is: (i) forming the L·T complex out of the capillary and injecting a small plug of the mixture into the capillary, (ii) separating free L from the L·T complex using a buffer free of T and L, and (iii) monitoring peaks corresponding to L and L·T. However, this method is not applicable to L·T complexes with relatively high values of koff (>10−2s−1) since the complex considerably decays during the separation (the typical separation time is ˜1000 s), which affects the accuracy of measurements. Many aptamers available today, especially those for small-molecule targets, have koff values that do not allow for their use as affinity probes in such CE-based analyses.
Screening for and selecting drug candidates and affinity probes. Finding new molecules capable of binding to therapeutic targets is essential for both drug discovery and development of new diagnostic methods. Target-binding molecules are used as potential drug candidates and as affinity probes for detecting targets. One of the most efficient ways of finding new target-binding molecules is the screening of complex biological (e.g. extracts from animal and plant tissues) and synthetic (e.g. combinatorial libraries of compounds) mixtures in binding assays (Chu et al., J. Org. Chem., 1993, 58, 648; Kuntz, Science 1992, 257, 1078; Baumbach et al., BioPhrm 1992, 5, 24; Pauwels et al., Nature, 1990, 343, 470; Sandler and Smith, Design of Enzyme Inhibitors as Drugs, 1989; Zuckermann et al., Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 4605; Fodor et al., Science, 1991, 251, 787; Lam et al., Nature 1991, 354, 81; Whitelegge et al., Am. J. Pharm. Genomics, 2001, 1, 29; Yao et al., J. Pharm. Sci., 2002, 91, 1923; Fortin and Nolan, Chemistry & Biology, 2002, 9, 670; Siegel, Current Topics in Medicinal Chemistry, 2002, 2, 13; Roberts, Xenobiotica, 2001, 31, 557; Gold, Nat. Biotechnol. 2002, 20, 671). Hereafter, the mixture to be screened will be called the sample.
Methods used to screen samples for target-binding molecules (screening methods) also fall into two broad categories: heterogeneous and homogeneous methods. In heterogeneous methods a target is affixed to a solid substrate such as, but not limited to, chromatographic support, beads, filters, walls of microtiter plates, or SPR sensor. The target can be affixed to the surface in a number of ways. The target can be chemically immobilized onto the surface or the target can be captured by anti-target antibodies, which are bound to the surface of the solid substrate. Also, if the target is a membrane protein, the target can be affixed to the surface indirectly through cells, which adhere to the surface. In heterogeneous analyses, the sample is incubated with the target. Molecules that bind to the target are captured on the surface. The non-bound components of the sample are washed out and the target-binding molecules can then be desorbed and analyzed. Heterogeneous screening methods have a number of drawbacks typical of heterogeneous analyses in general. The most serious is that affixing the target to the surface changes the structure of the target. The extent of such change will depend on the method of immobilization. The change in the structure can potentially affect binding of molecules to the target. This problem is especially severe when target-binding molecules recognize the structure of a large part of the target. In addition, the immobilization of the target on the surface may be time-consuming and expensive. Furthermore, non-specific binding to the surface leads to the “contamination” of ligands with non-ligands. In addition all existing heterogeneous screening methods share a very serious limitation. They do not provide a means of screening for target-binding molecules with specified ranges of binding parameters: kon, koff, and Kd. Optimum binding parameters will change from application to application. For example, the kon and koff values of a drug will influence its pharmacokinetics. Depending on the mechanism of a drug's action, the mechanism of its side effects, and the desirable regime of its administration, different binding parameters will be optimal (Bruice and Kahn, Curr. Opin. Chem. Biol., 2000, 4, 540; Van Oss, J. Mol. Recogn., 1997, 10, 203; Schuck, Curr. Opin. Biotechnol, 1997, 8, 498; White et al., Biochemistry, 1988, 27, 91222; Paton and Rang, Adv. Drug. Res. 1966, 3, 57). Different binding parameters can also be optimal for ligands to be used as affinity probes in different analyses. For example, in separation-based affinity analyses, it is desirable that koff be low to minimize complex decay during separation (German et al. Anal. Chem. 1998, 70, 4540; Berezovski and Krylov, Anal. Chem. 2003, 75, 1382). In fast clinical analyses, in contrast, it is essential that kon values be high to facilitate fast complex formation (Van Regenmortel et al., Immunological Investigations, 1997, 26, 67; Krishnan et al., Env. Health Perspectives, 1994, 102). Other types of analyses may require specific ranges of Kd. Thus, it would be very beneficial to have a means of selecting molecules with desirable ranges of kon, koff, and Kd.
In homogeneous screening methods, the target is mixed with the sample in solution and then subjected to either electrophoresis or chromatography. The free target is separated from the target-ligand complex based on differences in their chromatographic or electrophoretic properties. The major advantage of homogeneous screening methods over heterogeneous ones, is that they screen for molecules that bind to targets with an unmodified structure.
Despite the well-known advantages of CE, only a few CE-based homogeneous screening methods are available. Patent WO 97/22000 describes the use of a CE-based homogeneous screening method to detect compounds present in natural samples that could complex with a known target, as a tool for identifying potential therapeutic or diagnostic compounds. The method monitored changes in the migration pattern of the target during electrophoresis as a sentinel of complex formation. Due to this requirement, the method is limited to detectable concentrations of target and ligand. In addition, this method is limited to detecting complexes that remain bound as they migrate past the detector. These ligands can be referred to as tight binding ligands (TBL). Moderate (MBL) and weak binding ligands (WBL) dissociate before reaching the detector and do not produce a detectable shift in the migration pattern of the target. Hence, they are not detected.
Methods using tight binding competitive ligands (TBCL) were designed to overcome this problem. U.S. Pat. No. 6,299,747 and WO 00/79260 disclose methods of detecting new therapeutic regulatory and diagnostic compounds in complex biological materials using a known competitive ligand to the target. The TBCL is added to the target/sample mixture and peak changes of the unbound target or the target associated with the TBCL are monitored alone or together. MBLs and WBLs in the sample mixture that result in an increased unbound peak or a decreased target/TBCL peak, are detected. The patent indicates that MBLs and TBLs are detectable in the picomolar (pM) to low nanomolar (nM) range. This method, however, requires that a known TBCL exist and be available, and is limited to embodiments using detectable concentrations of target since the method relies on tracking changes in the migration pattern of the target. Homogeneous methods employing CE have also been combined with analytical methods to aid in the identification of target binding ligands. WO 00/03240 describes a method where the CE technique is combined with mass spectrometry, for screening complex samples. Again this method relies on complexes that migrate stably through the CE instrument.
Furthermore attempts have been made to select ligands that bind the chosen target with a selected binding strength. WO 99/34203 provides a method for determining relative binding strengths of ligands that bind the chosen target. This permits ligands to be ranked according to their relative binding strengths, which can aid in prioritizing further analysis. This method, however, does not provide a means of obtaining real binding parameters as the terms “affinity” and “relative affinities” are used in this patent “in a general sense, and do not necessarily refer to a hit compound's “binding affinity” to a target in an equilibrium situation”.
To conclude, the existing CE-based homogeneous screening methods (WO 97/22000, U.S. Pat. No. 6,299,747, WO 00/79260 and WO 00/03240) share several serious disadvantages. They all require: (i) detectable amounts of target and (ii) large amounts of ligands capable of inducing detectable shifts of the peak of target. Due to the first requirement they are unable to select ligands for targets present in small amounts and ligands with specified binding parameters. The second requirement makes it impossible to screen for ligands that constitute only a very small portion of the screened sample, such as components of a large combinatorial library where ligand representation is low. Indeed, the relative amount of ligands, which have required binding parameters in combinatorial libraries can be as low as 10−13 (Gold, J. Biol. Chem. 270, 1995, 13581) such that the target-ligand complexes will be present in undetectable amounts, and will not introduce a mobility shift to the target. Finally, the existing CE-based homogeneous screening methods do not provide a means of selecting ligands with a specified range of kon, koff, and Kd parameters.